A rapid proliferation of data-intensive and autonomous applications is redefining the limits of modern computing systems, as the growth of computational scale and complexity outpaces the performance gains achievable through hardware scaling alone. This widening gap is driving a shift toward increasingly integrated and scalable heterogeneous architectures, in which diverse, specialized chips operate in coordinated and efficient ways. However, programming and designing these complex systems remain difficult, time-consuming, and prone to errors. This project addresses this critical challenge by developing a unified and easy-to-use framework that empowers domain experts to design and program these advanced computing systems efficiently. The resulting advances will support real-world applications such as autonomous physical systems, adaptive intelligent agents, and advanced healthcare technologies, enabling more efficient, responsive, and accessible computing capabilities. The project will create open-source tools, datasets, and benchmarks that will be shared through workshops, tutorials, and demonstrations. The research will be integrated into new curricular materials, support undergraduate mentorship and graduate training through hands-on, community-shared tools, and strengthen workforce preparation in microelectronics, computer science, and engineering. The project will establish a scalable high-level synthesis and programming framework that enables verified and efficient integration of heterogeneous computing systems across diverse hardware backends. This project consists of three tightly integrated research thrusts. First, it will construct a multi-level intermediate representation infrastructure that supports extensible compilation across field-programmable gate arrays, graphics processing units, tensor accelerators, processing-in-memory architectures, and application-specific integrated circuits, enabling reusable abstractions and flexible backend targeting. S